sensors

Article
Traffic Management for Emergency Vehicle Priority
Based on Visual Sensing
Kapileswar Nellore 1, * and Gerhard P. Hancke 1,2
1 Department of Electrical, Electronic and Computer Engineering, University of Pretoria, Pretoria 0028,
South Africa; ghancke@ieee.org
2 Department of Computer Science, City University of Hong Kong, Hong Kong, China
* Correspondence: u15342396@tuks.co.za; Tel.: +27-61-859-5390; Fax: +27-12-362-5000

Academic Editor: Paolo Bellavista

Received: 16 August 2016; Accepted: 25 October 2016; Published: 10 November 2016

Abstract: Vehicular traffic is endlessly increasing everywhere in the world and can cause terrible
traffic congestion at intersections. Most of the traffic lights today feature a fixed green light sequence,
therefore the green light sequence is determined without taking the presence of the emergency
vehicles into account. Therefore, emergency vehicles such as ambulances, police cars, fire engines,
etc. stuck in a traffic jam and delayed in reaching their destination can lead to loss of property
and valuable lives. This paper presents an approach to schedule emergency vehicles in traffic.
The approach combines the measurement of the distance between the emergency vehicle and an
intersection using visual sensing methods, vehicle counting and time sensitive alert transmission
within the sensor network. The distance between the emergency vehicle and the intersection is
calculated for comparison using Euclidean distance, Manhattan distance and Canberra distance
techniques. The experimental results have shown that the Euclidean distance outperforms other
distance measurement techniques. Along with visual sensing techniques to collect emergency vehicle
information, it is very important to have a Medium Access Control (MAC) protocol to deliver the
emergency vehicle information to the Traffic Management Center (TMC) with less delay. Then only
the emergency vehicle is quickly served and can reach the destination in time. In this paper, we have
also investigated the MAC layer in WSNs to prioritize the emergency vehicle data and to reduce the
transmission delay for emergency messages. We have modified the medium access procedure used
in standard IEEE 802.11p with PE-MAC protocol, which is a new back off selection and contention
window adjustment scheme to achieve low broadcast delay for emergency messages. A VANET model
for the UTMS is developed and simulated in NS-2. The performance of the standard IEEE 802.11p
and the proposed PE-MAC is analysed in detail. The NS-2 simulation results have shown that
the PE-MAC outperforms the IEEE 802.11p in terms of average end-to-end delay, throughput and
energy consumption. The performance evaluation results have proven that the proposed PE-MAC
prioritizes the emergency vehicle data and delivers the emergency messages to the TMC with
less delay compared to the IEEE 802.11p. The transmission delay of the proposed PE-MAC is also
compared with the standard IEEE 802.15.4, and Enhanced Back-off Selection scheme for IEEE 802.15.4
protocol [EBSS, an existing protocol to ensure fast transmission of the detected events on the road
towards the TMC] and the comparative results have proven the effectiveness of the PE-MAC over
them. Furthermore, this research work will provide an insight into the design of an intelligent urban
traffic management system for the effective management of emergency vehicles and will help to save
lives and property.

Keywords: VANETs; audio visual sensing; emergency vehicle; traffic lights; traffic monitoring;
priority; distance measurement techniques

Sensors 2016, 16, 1892; doi:10.3390/s16111892 www.mdpi.com/journal/sensors

Sensors 2016, 16, 1892 2 of 22

Sensors 2016, 16, 1892 2 of 22

1. Introduction
1. Introduction
TheThe
traffic light
traffic control
light control plays
playsa avital
vitalrole
role in
in any intelligenttraffic
any intelligent trafficmanagement
management system.
system. TheThe
greengreen
lightlight
sequence
sequence and green light duration are the two key aspects to be considered in traffic lightlight
and green light duration are the two key aspects to be considered in traffic
control. In many
control. In many countries,
countries, most
mosttraffic
trafficlights
lights feature fixedsequences
feature fixed sequencesand and light
light length
length duration.
duration. FixedFixed
control methods
control methods arearehowever
howeveronly onlysuitable
suitable for
for stable andregular
stable and regulartraffic,
traffic,butbut
notnot
forfor dynamic
dynamic traffic
traffic
situations. Looking
situations. Looking at at
the thepresent
presentstate
stateofof practice, the green
practice, the greenlight
lightsequence
sequence is is determined
determined without
without
taking the possible presence of emergency vehicles into account. Therefore, emergency
taking the possible presence of emergency vehicles into account. Therefore, emergency vehicles such as vehicles such
as ambulances,
ambulances, policepolice
cars,cars,
fire fire engines,
engines, etc.etc. mustwait
must waitinintraffic
traffic at
at an
anintersection
intersection asasdepicted in Figure
depicted 1 1
in Figure
whichwhich delays
delays theirtheir arrival
arrival at their
at their destination
destination causing
causing loss loss of lives
of lives and and property.
property. In Ireland,
In Ireland, an
an average
average
of 700 of 700
fatalities wasfatalities
noted waseverynoted
yearevery
due year due
to late to late ambulance
ambulance responses responses
[1]. [1].

Figure 1.1.Emergency
Figure Emergency vehicle waitingatatan
vehicle waiting anintersection.
intersection.

An increased
An increased volume
volume of vehicles
of vehicles notnot only
only increasesthe
increases theresponse
responsetimestimesofofemergency
emergency vehicles,
vehicles, but
but it also increases the chances for them being involved in accidents. The emergency vehicle entering
it also increases the chances for them being involved in accidents. The emergency vehicle entering
an intersection at a high speed on a red light poses danger to traffic on other roads and can cause
an intersection at a high speed on a red light poses danger to traffic on other roads and can cause
accidents. The National Highway Traffic Safety Administration (NHTSA) has collected the ground
accidents. The National Highway Traffic Safety Administration (NHTSA) has collected the ground
ambulance crash data for the U.S between 1992 and 2011 (20 years) [2]. There were an estimated
ambulance crash of
annual mean data formotor
4500 the U.S between
vehicle 1992
traffic and 2011
crashes (20 years)
and 1500 injury[2]. There
crashes were anan
involving estimated annual
ambulance.
mean of 4500 motor vehicle traffic crashes and 1500 injury crashes involving an
In 20 years (1992–2011), 662 persons were killed and 52,000 persons were estimated to be injured in ambulance. In 20 years
(1992–2011), 662 persons
such crashes, including were killed and
ambulance 52,000passengers,
drivers, persons were estimated toand
non-occupants be injured
occupants in such crashes,
of other
including
vehicles.ambulance
From thedrivers,
statisticspassengers,
of emergency non-occupants
vehicle accidentsand occupants
in U.S, there of other
were vehicles. From the
roughly 31,600
accidents
statistics involving fire
of emergency vehicles
vehicle resultedin
accidents inU.S,
645 fatalities
there were overroughly
a 10 year31,600
periodaccidents
(2000–2009) and 300 fire
involving
fatalities
vehicles that occur
resulted in 645 every year during
fatalities over apolice
10 yearpursuits
period[3]. From the current
(2000–2009) and 300 problem section,
fatalities that it can be
occur every
yearunderstood
during police that,pursuits
there is a[3].
serious
Fromneedthe for an intelligent
current problemtraffic
section,management system for the
it can be understood effective
that, there is a
management
serious need for of anboth the normal
intelligent andmanagement
traffic emergency vehicles.
system for the effective management of both the
Many researchers have developed pre-emption systems that utilize the distance between the
normal and emergency vehicles.
emergency vehicle and the intersection based on GPS for the signaling time. In this paper, we present
Many researchers have developed pre-emption systems that utilize the distance between the
a new approach to calculate the distance between the emergency vehicle and intersection using a
emergency vehicle and the intersection based on GPS for the signaling time. In this paper, we present
real-time video feed from the cameras at intersections. An intelligent algorithm to shorten the travel
a newtimeapproach to calculate
by scheduling both thethe distance
green betweenand
light sequence thegreen
emergency vehiclebased
light duration and intersection
on the measuredusing a
real-time video
distance feedpresented.
is also from the cameras
A MACatprotocol
intersections.
named An intelligent
PE-MAC algorithm
is also proposed to shorten
to deliverthethe
travel
timeemergency
by schedulingvehicle both the greentolight
information the sequence
TMC withand less green
delay.light duration
A VANET based
model for on
thethe measured
UTMS is
distance is also presented. A MAC protocol named PE-MAC is also proposed
developed, the NS is configured according to the proposed protocol and simulated in NS-2. We have to deliver the emergency
vehicle information
planned to analyseto the
andTMC comparewiththe
lessperformance
delay. A VANETof the model
PE-MAC forwith
the UTMS is developed,
the standard the NS is
IEEE 802.11p
and other
configured existing MAC
according to the protocols.
proposed protocol and simulated in NS-2. We have planned to analyse
and compare the performance of the PE-MAC with the standard IEEE 802.11p and other existing
MAC protocols.
Sensors 2016, 16, 1892 3 of 22

The remaining parts of this paper are structured as follows: Section 2 provides a review of
related work. The UTMS architecture, functionality of TMC, distance measurement techniques and the
proposed distance based algorithm are explained in detail in Section 3. The proposed PE-MAC and
VANET model is presented in detail in Section 4. Experimental results are presented and discussed in
Section 5. Section 6 concludes the paper.

2. Related Work
With an increasing amount of vehicles on the road, traffic congestion and transportation delays
are increasing worldwide. Emergency vehicles, such as ambulances, fire engines and police cars,
should be capable to react to emergency calls with minimum delay. The excellence of the emergency
service depends on how fast the emergency vehicles can reach the incident location. If the emergency
vehicle gets stuck in a traffic jam and its arrival at the incident location is delayed it can cause loss of
lives and property. There is a need for smart traffic management systems based on priority and traffic
density to improve the transportation efficiency and response times of emergency services.
Extensive work has been conducted on how to efficiently use traffic information to determine
green light sequences. A number of traffic management schemes have been implemented to prioritize
emergency vehicles [4]. Most of this research is associated with intelligent traffic control system
design for providing clearance for emergency vehicles [5–9]. The traffic conditions are measured using
cameras and the traffic parameters are estimated using the lane center edges [10]. An area-based image
processing technique for the detection of traffic density was presented in [11].
Traditional traffic systems include strobe emitters, or light emitters that notice problems such as
blocked line-of-sight and excessive noise [12]. Recent technologies such as infrared (IR) and Global
Positioning System (GPS) have been used to detect emergency vehicle presence and calculate the
real-time traffic density [13]. RFID tags have been used to identify the presence of emergency vehicles
and the inductive loop method is used to count vehicles [14]. The widely used traffic detection methods,
include video-based detection, microwave detection, radar detection, ultrasonic detection, etc. [15,16].
Wireless Sensor Networks (WSNs) allow for embedded sensors to be interconnected for observing
and controlling consumer and industrial actions [17,18]. The use of Vehicular Sensor Networks (VSNs)
or infrastructure WSNs have been proved to be promising solutions for monitoring and management
of traffic. WSNs are flexible in terms of and energy efficiency and data collection type, e.g., video [19].
If a vehicle contains a WSN node, then localization algorithms [20–22] can be used to determine its
location. The General Packet Radio Service (GPRS) technology can also be used for the dynamic control
of traffic signals [23].
Nowadays, there are many emergency vehicle pre-emption (EVP) system designs including strobe
light systems, infrared emitters, acoustic systems, and radio-based emitter/detector systems [24].
Usually, pre-emption works on the principle that an emergency vehicle is identified by sensors at
each intersection and the traffic light controller switches ON the green light and holds it until the
emergency vehicle exits the intersection [25]. The info-gap decision theory can also be used for actor
coordination [26].
The green wave system, which provides a green wave to emergency vehicle by switching ON
the green lights in the path of the emergency vehicle can be found in [27]. If the wave is disturbed,
the vehicle queue in a green wave shoots up and may lead to over-saturation [28,29]. Technologies
like RFID, Zig Bee, and Global System for Mobile communication (GSM) can be used for designing an
intelligent traffic control system [30].
An RFID- and GPS-based automatic lane clearance system for ambulances is proposed in [31].
The main goal of this system is to reduce the travel time of an ambulance to the hospital by
automatically clearing the lane, before the ambulance reaches the intersection. Vehicular Ad-Hoc
Networks (VANETs) have been proven effective communication methods between an emergency
vehicle and a traffic control system [32]. The connection admission control (CAC) algorithm has proven
its better QOS and complexity performance [33]. Fuzzy control approaches have been adapted to
monitor real-time traffic and handle the crowded traffic flow [34].
Sensors 2016, 16, 1892 4 of 22

From the previous discussion, it can be understood that there are many techniques for providing
clearance for emergency vehicles. Each of these techniques seems to have merits and demerits. In this
research work an approach that combines the measurement of distance between the emergency vehicle
and an intersection using visual sensing methods and PE-MAC for fast transmission of emergency
vehicle information is proposed.

3. Proposed Methodology
In this section, we present the architecture of the urban traffic management system considered in
our work, the functionality of the traffic management center, distance measurement techniques and a
distance-based emergency vehicle dispatching algorithm.

3.1. The Architecture of an Urban Traffic Management System (UTMS)

WSNs and VANETs for smart cities [35] are becoming a reality with increased options for
area coverage and connectivity stemming from machine-to-machine communication [36] and the
Internet-of-Things [37–41]. An Urban Traffic Management System (UTMS), depicted in Figure 2, refers
to a system that integrates sensing technologies, data processing techniques, wireless communications
Sensors 2016, 16, 1892 4 of 22
and advanced technologies to reduce traffic congestion, travel time, fuel consumption and provide
priority-based signaling. On obtaining the data of emergency vehicles from sensors, the Traffic
From the previous discussion, it can be understood that there are many techniques for providing
Management
clearance forCentre (TMC)
emergency follows
vehicles. Eachthe distance-based
of these emergency
techniques seems vehicleand
to have merits dispatching
demerits. In(DBEVD)
this
algorithm and provides signals to the emergency vehicle immediately.
research work an approach that combines the measurement of distance between the emergency The TMC of the present
intersection
vehicle and (TMCA) providesusing
an intersection the TMC
visualofsensing
next intersection
methods and(TMCB)
PE-MACwith
forthe velocity
fast of emergency
transmission of
vehicle and vehicle
emergency vehicle count moving
information towards intersection-B. As the TMCB knows the velocity of the
is proposed.
emergency vehicle, it can estimate its arrival time at intersection-B. The TMCB determines the green
3. Proposed
light sequenceMethodology
and green light duration based on the estimated arrival time of the emergency vehicle,
and theInvehicle countwe
this section, value sentthe
present by architecture
the TMCA. of Therefore,
the urbanthe emergency
traffic vehicles
management systempasses through the
considered
intersections with no or little delay. On using the proposed algorithm, the UTMS can
in our work, the functionality of the traffic management center, distance measurement techniques effectively handle
theand a distance-based
emergency vehiclesemergency
and thus savevehicle dispatching
lives algorithm.
and property.

Figure 2.
Figure 2. Architecture
Architectureofofan
anurban
urbantraffic management
traffic system.
management system.

3.1. The Architecture of an Urban Traffic Management System (UTMS)

3.2. Traffic Management Center
WSNs and VANETs for smart cities [35] are becoming a reality with increased options for area
In this subsection, we describe the functionality of the Traffic Management Centre (TMC) and
coverage and connectivity stemming from machine-to-machine communication [36] and the Internet-
theof-Things
role of each of itsAn
[37–41]. units.
UrbanTheTraffic
schematic of a typical
Management Traffic
System Management
(UTMS), depicted Centre (TMC)
in Figure is presented
2, refers to a
in Figure 3. The presently used traffic light pre-emption systems can be categorised based
system that integrates sensing technologies, data processing techniques, wireless communications on their
and advanced technologies to reduce traffic congestion, travel time, fuel consumption and provide
priority-based signaling. On obtaining the data of emergency vehicles from sensors, the Traffic
Management Centre (TMC) follows the distance-based emergency vehicle dispatching (DBEVD)
algorithm and provides signals to the emergency vehicle immediately. The TMC of the present
intersection (TMCA) provides the TMC of next intersection (TMCB) with the velocity of emergency
 Sensors 2016, 16, 1892 5 of 22
Sensors 2016, 16, 1892 5 of 22

operation as:
operation as:optical
opticalsystems,
systems,radio-controlled
radio-controlled systems,
systems, GPS-based
GPS-based systems
systems and acoustic
and acoustic sensor-
sensor-based
based systems.
systems. AcousticAcoustic sensor-based
sensor-based systemssystems outperform
outperform the otherthe other pre-emption
pre-emption systems
systems in in accuracy
terms of terms of
accuracy and installation cost. Extensive work has been conducted on detecting emergency
and installation cost. Extensive work has been conducted on detecting emergency vehicles based vehicles
on
based on their siren sounds. We summarize the proposed approaches for emergency
their siren sounds. We summarize the proposed approaches for emergency vehicle detection based vehicle
on
detection
siren based
sounds on siren
in Table 1. sounds in Table 1.

Figure 3. Schematic of a traffic management centre.

Figure 3. Schematic of a traffic management centre.

Table 1.
Table Approaches for
1. Approaches for emergency
emergency vehicle
vehicle detection
detection based
based on
on siren
siren sounds.
sounds.

Ref. Ref. Proposed Approach

Proposed Approach Outcome
Outcome Comments
Comments
Design and implementation of The traffic light sequence is interrupted
Design and implementation of The traffic light sequence is interrupted with the
acoustic sensor-based Detects theDetects
emergency vehicle by
the emergency with the approaching emergency
[42] [42] acoustic sensor-based automatic approaching emergency vehicle and thereby its
automatic traffic control their sirenvehicle
sounds.by their siren sounds. vehicle and thereby its waiting time at
traffic control system. waiting
system. the time at the intersection
intersection is reduced.is reduced.
Low computational Designed a siren sound The proposed method outperforms the
Low computational
[43] microcontroller-based siren Designed
sound adetection
siren sound detection
system with low The proposed method
existing siren outperforms
detection methodsthein
existing
[43] microcontroller-based
detection system. siren system with low processing
processing power. sirenterms
detection methods inpower
of processing terms and
of processing
cost.
sound detection system. power. power and cost.
Detects the siren sound in
This work only detects the ambulance
Detection of siren sounds using 0 dB (S/N ratio). Detects the
[44] Detects the siren sound in 0 dB This siren
work sound and neither
only detects alerts thesiren
the ambulance traffic
sound
FastofFourier
Detection Transform
siren sounds using(FFT) siren sound using the
[44] (S/N ratio).Doppler
Detects effect.
the siren nor changes
and neither alerts the traffic nor
signals.
changes the traffic
Fast Fourier Transform (FFT)
sound using the Doppler effect. signals.
Capable of detecting the The proposed algorithm outperforms
Detection of siren sounds based emergency vehicle in the the complex pattern recognition
[45]
on a pitch detection algorithm.
Capable ofpresence
detectingofthe
pitched and The proposed
algorithms. The siren
algorithm signal missthe
outperforms rate of
complex
Detection of siren sounds based
[45] emergencynon-pitched noise.
vehicle in the presence therecognition
pattern algorithm is very low. The siren signal
algorithms.
on a pitch detection algorithm.
of pitched and non-pitched noise. miss The
rate proposed
of the algorithm
systemisalert‘s
very low.
the drivers
Emergency vehicle’s siren and Cost effective solution.
of normal vehicles and pedestrians
[46] flashing light detection based on Distinct emergency vehicles
about the approaching emergency
acoustic
Emergency and optical
vehicle’s sensors.
siren and are detected. The proposed
Cost effective solution. Distinct vehicle. system alert‘s the drivers of normal
[46] flashing light detection based vehicles and pedestrians about the approaching
emergency vehicles are detected. The proposed system for source
on acoustic and optical sensors. emergency vehicle.
Cross microphone array-based Determines the incoming detection outperforms the existing
[47]
emergency vehicle detection. direction of siren sound. sound intensity techniques. It delivers
The proposed system for source detection
precise warning data to the driver.
Cross microphone array- based Determines the incoming outperforms the existing sound intensity
[47] Analyses and detects the techniques.
The proposed work alertswarning
the driver
emergency vehicle
Digital imagedetection.
sensor-based direction of siren sound. It delivers precise data to
emergency vehicle in an the driver.when an emergency vehicle is detected.
[48] emergency vehicle detection and
image using image It is not cost effective as it needs the
display system for a vehicle.
processing techniques. cameras to be mounted on the vehicle.
Digital image sensor-based Analyses and detects the The proposed work alerts the driver when an
emergency vehicle detection emergency vehicle in an image emergency vehicle is detected. It is not cost
[48]
and display system for a using image processing effective as it needs the cameras to be mounted on
The acoustic sensors collect the siren signals and forward them to the Road Side Unit (RSU).
vehicle. techniques. the vehicle.
The Road Side Unit (RSU) includes a frequency measuring controller (Arduino UNO) to detect the
emergency vehicles. The RSU collects the siren signals from the acoustic sensors and forwards them
The acoustic sensors collect the siren signals and forward them to the Road Side Unit (RSU). The
Road Side Unit (RSU) includes a frequency measuring controller (Arduino UNO) to detect the
Sensors 2016, 16, 1892 6 of 22

to the frequency measuring controller. The controller detects the emergency vehicle by its siren
frequencies. The controller measures the frequencies of siren signals and computes the average of
measured frequencies. The frequency measuring controller sends the alert signal to the traffic signal
controller (Arduino Mega), if the frequency is between the range of yelp or wail. The traffic signal
controller stops the fixed sequence and light length algorithm and executes the emergency vehicle
dispatching algorithm on receipt of arriving emergency vehicle information. The data collection
module gathers the data from all the RSU’s and forwards it to Traffic Signal Control Module (TSCM).
The TSCM has two units, namely traffic analysis unit and traffic signal controller (Adriano mega).
The camera sensor captures the real-time traffic video and inputs the traffic analysis unit, where the raw
traffic data is processed and analyzed. The traffic controller unit gets the data like distance, velocity,
traffic density, vehicle count, etc. from the traffic analysis unit. The controller executes the proposed
algorithm and sends its decision to traffic lights. After the passage of an emergency vehicle, the system
resumes its normal operation, i.e., fixed sequence and light length algorithm. In the following, we
discuss distance measurement techniques, vehicle counting methods, a distance-based emergency
vehicle dispatching algorithm and the simulation environment.

3.3. Distance Measurement Techniques

At present, the popular distance measurement techniques include ultrasonic, infrared, laser,
machine vision and radar measurements. The distance measurement based on machine vision obtains
the value of the distance by the real-time processing of visual signals. There are different techniques to
measure the distance between the vehicle and the camera. We perform the distance measurement by
computing the Euclidian distance, Manhattan distance and Canberra distance [49,50]. In the following
we briefly discuss these distances:

3.3.1. Euclidean Distance

It is the geometric distance in the multidimensional space. One technique that may suit a wide
variety of image analysis applications is the distance transform or a Euclidean distance map. In general,
if a = (a1 , a2 , a3 , . . . , am ) and b = (b1 , b2 , b3 , . . . , bm ) are the two points in m-space, then the Euclidean
distance (dECD ) between points a and b or b and a is as follows:
m
d ECD ( a, b) = (b1 − a1 )2 + (b2 − a2 )2 + . . . . . . . . . +(bm − am )2 = ∑ ( bi − a i ) 2 (1)
i =1

Let the pixels within a two-dimensional frame (x, y) be divided into two classes: object pixels and
background pixels. For 2D points, a = (a1 , a2 ) and b = (b1 , b2 ) the Euclidean distance is:

d ECD ( a, b) = (b1 − a1 )2 + (b2 − a2 )2 (2)

3.3.2. Manhattan Distance

The Manhattan distance between two points is the absolute sum of the differences of their
coordinates. In general, if a = (a1 , a2 , a3 , . . . , am ) and b = (b1 , b2 , b3 , . . . , bm ) are the two points in
m-space, then the Manhattan distance (dMHD ) between a and b is defined as follows:
m
d MHD ( a, b) = ∑ | a i − bi | (3)
i =1

The Manhattan distance between points a = (a1 , a2 ) and b = (b1 , b2 ) is:

d MHD ( a, b) =| a1 − b1 |+| a2 − b2 | (4)

Sensors 2016, 16, 1892 7 of 22

3.3.3. Canberra Distance

This distance is a weighted version of the Manhattan distance and is frequently used for data
scattered around an origin. In general, if a = (a1 , a2 , a3 , . . . , am ) and b = (b1 , b2 , b3 , . . . , bm ) are the
two points in m-dimensional real vector space, the Canberra distance (dCAD ) between a and b is given
as follows:
m
| a − bi |
dCAD ( a, b) = ∑ i (5)
| a |+|bi |
i =1 i

The Canberra distance (dCAD ) between two vectors a, b in 2D real vector space is given by:

| a1 − b1 | | a − b2 |
dCAD ( a, b) = + 2 (6)
| a1 |+|b1 | | a2 |+|b2 |

Since sensors’ performance in distance calculation is highly dependent on the environmental

conditions, vision systems are highly preferred for moving vehicle distance calculations. The accuracy
of the distance measurement techniques is computed by comparing the simulation results with the
true practically measured distance.

3.4. Vehicle Counting Method

An input video sequence of road traffic can be processed and analyzed to get vehicle counts and
speeds. The traffic management center can utilize this information in a traffic signal control module,
resulting in an efficient traffic management. The vehicle counting method includes the following steps:

i. Grayscale conversion.
ii. Foreground extraction.
iii. Defining Region of Interest (ROI).
iv. Thresholding.
v. Filling the holes.
vi. Morphological operations.
vii. Detect the vehicles entering the ROI.
viii. Count the vehicles.

3.5. Distance Based Emergency Vehicle Dispatching Algorithm

In order to achieve optimal traffic light control to provide clearance for any emergency vehicle
and to shorten its travel time, we propose a distance-based emergency vehicle dispatching algorithm.
We assumed only one emergency vehicle per direction. The proposed algorithm is represented in
Figure 4. The EVs in the flowchart represent the emergency vehicles. The proposed algorithm has
mainly six steps:

1. The sensor senses the presence of emergency vehicles. The emergency vehicles are ambulances,
fire engines and police cars.
2. Calculate the distance between the emergency vehicle and the intersection.
3. The controller checks that the arriving emergency vehicles are at the same distance or not. If they
are at the same distance, the controller randomly chooses the direction to set the green light.
Else, he chooses the direction set in ascending order with the distance.
4. Determine the green light duration based on the measured distance values and send these values
to the traffic lights.
5. Verify the passage of the emergency vehicle and measure the speed of the emergency vehicle
and count the vehicles moving along with the emergency vehicle towards next intersection.
The system sends the measured data to the next intersection.
6. The controller checks for the presence of the emergency vehicle. If no vehicle, then it resumes
normal operation. Else, it continues repeats from step 2 to step 6.
Sensors 2016, 16, 1892 8 of 22
Sensors 2016, 16, 1892 8 of 22

Figure4.4.Distance-based
Figure Distance-based emergency
emergencyvehicle
vehicledispatching
dispatchingalgorithm.
algorithm.

In this research work, we focused on visual sensing methods for determining green light
In this research work, we focused on visual sensing methods for determining green light sequences
sequences and green light duration. Distance measurement techniques help us to find the nearest
andemergency
green light duration.
vehicle to the Distance measurement
intersection and determine techniques helpsequence.
the green light us to find the nearest emergency
vehicle Up
to the intersection and determine the green light sequence.
to now, we have discussed in detail visual sensing methods to collect the emergency vehicle
Up to now,Itwe
information. have
is also discussed
important to in detail
talk aboutvisual sensing
how fast methodsinformation
the measured to collect the emergency
is delivered to vehicle
the
information.
TMC. For thatIt iswealso important
have to talk
investigated theabout
Mediumhow fast the
Access measured
Control (MAC)information is delivered
layer in sensor networks to to the
TMC. For that
prioritize thewe have investigated
emergency vehicle datathe
andMedium
to reduceAccess Control (MAC)
the transmission layer
delay for in sensormessages.
emergency networks to
We have
prioritize themodified
emergency the medium access
vehicle data procedure
and to reduce used
the in the IEEE 802.11p
transmission delay standard with the
for emergency PE-
messages.
WeMAC protocol, which
have modified is a new
the medium back-off
access selection
procedure and
used in contention window
the IEEE 802.11p adjustment
standard withscheme to
the PE-MAC
achievewhich
protocol, low broadcast
is a newdelays forselection
back-off emergency andmessages.
contentionThe window
PE-MAC adjustment
is explained scheme
in detail to
in achieve
the next low
section. delays for emergency messages. The PE-MAC is explained in detail in the next section.
broadcast

4. 4. PE-MACProtocol
PE-MAC Protocol

In In this section, we present the Priority for Emergency Traffic based MAC (PE-MAC) Protocol,
this section, we present the Priority for Emergency Traffic based MAC (PE-MAC) Protocol,
PE-MAC algorithm and the developed VANET model.
PE-MAC algorithm and the developed VANET model.

4.1. Priority for Emergency Traffic Based MAC (PE-MAC) Protocol

The IEEE 802.11p protocol has been developed by the IEEE 1609 working group as a key
communication standard for vehicular networking. The IEEE 802.11p basically adopts the carrier sense
 Sensors 2016, 16, 1892 9 of 22
Sensors 2016, 16, 1892 9 of 22
4.1. Priority for Emergency Traffic Based MAC (PE-MAC) Protocol
The IEEE 802.11p protocol has been developed by the IEEE 1609 working group as a key
multiple access with collision avoidance (CSMA/CA) with exponential back-off mechanism for packet
communication standard for vehicular networking. The IEEE 802.11p basically adopts the carrier
access control. When a station wants to send a packet, first it has to listen to the channel, which is
sense multiple access with collision avoidance (CSMA/CA) with exponential back-off mechanism for
referred to as the carrier sensing. If the channel is free for a time known as the distributed inter frame
packet access control. When a station wants to send a packet, first it has to listen to the channel, which
spacingis referredtime,
(DIFS) to as the
the station will transmit
carrier sensing. If the a requestis to
channel send
free for (RTS)
a time to the destination.
known The destination
as the distributed inter
will respond with a clear to send (CTS) if it is available to receive data. When the
frame spacing (DIFS) time, the station will transmit a request to send (RTS) to the destination. source stationThe
receives
the CTS, it will transmit
destination its data.
will respond with aThe network
clear allocation
to send (CTS) if it isvector (NAV)
available indicates
to receive data. the
Whentime
theamount
source the
channelstation receives
is busy. All the
theCTS, it willsent
packets transmit
in theitsnetwork
data. Thehold
network
thisallocation vector (NAV)
NAV information. indicates
After the has
the data
time amount the channel is busy. All the packets sent in the network hold
been correctly received at the destination station, it will send an acknowledgment (ACK) back to the this NAV information.
senderAfter the data has been correctly received at the destination station, it will send an acknowledgment
station. At this point, if the sender has more data to transmit, it will again begin its back-off and
(ACK) back to the sender station. At this point, if the sender has more data to transmit, it will again
repeat the process. The frame structure and the CSMA/CA process are demonstrated in Figure 5a,b,
begin its back-off and repeat the process. The frame structure and the CSMA/CA process are
respectively. The short inter frame spacing (SIFS) is used as the wait time between the RTS, CTS, DATA
demonstrated in Figures 5a, b, respectively. The short inter frame spacing (SIFS) is used as the wait
and ACKtime betweenThe
frames. SIFS CTS,
the RTS, ensuresDATAthatandtheACK
other node The
frames. doesSIFS
notensures
wrongly determine
that the other that
nodethe channel
does not is
idle during the handshake.
wrongly determine that the channel is idle during the handshake.

(a)

(b)
Figure 5. IEEE 802.11p (a) Superframe structure; (b) CSMA/CA process.
Figure 5. IEEE 802.11p (a) Superframe structure; (b) CSMA/CA process.

The main goal of our work is to provide priority to the emergency vehicle messages and to reduce
their broadcast delay. To achieve this goal, we have planned to modify the medium access procedure
used in IEEE 802.11p standard with the PE-MAC protocol, which is a new back-off selection and
contention window adjustment scheme to achieve low broadcast delays. We have four different types
of data: (1) ambulance data; (2) fire engine data; (3) police car data; and (4) normal vehicle data.
The priority assignment to these four different types of data is given in Table 2.
different types of data: (1) ambulance data; (2) fire engine data; (3) police car data; and (4) normal
vehicle data. The priority assignment to these four different types of data is given in Table 2.

Table 2. Data types with priority assignment and access requirements.

Sensors 2016, 16, 1892 10 of 22
Data Type, Index Priority Assigned Back-off Values Medium Access Requirement

Table1 2. Data
Ambulance data, Firsttypes with(Highest)
priority priority assignment and access requirements.
BOFF1 Fast

Firefighter
Data data, 2
Type, Index Second
Priority priority
Assigned BOFF2 > BValues
Back-off OFF1 Fast Requirement
Medium Access
Ambulance data, 1 First priority (Highest) B Fast
Police car data, 3 Third priority BOFF3 OFF1
> BOFF2 Fast
Firefighter data, 2 Second priority BOFF2 > BOFF1 Fast
Police car data, 3 Third priority BOFF3 > BOFF2 Fast
Normal
Normal vehicle
vehicle data,
data, 4 4 Fourth
Fourthpriority
priority (least)
(least) BOFF4 >>BBOFF3
B Fast or or
Fast slow
slow
OFF4 OFF3

We have
We have assumed
assumed aa random
random variable
variable “Y”,
‘Y’, which
which represents
represents the back-off delay
the back-off delay value
value in
in the
the
contention window interval [0, CW]. The back-off delay value (Y) can be defined
contention window interval [0, CW]. The back-off delay value (Y) can be defined as:as:
Y/T = 1 ~ Normal distribution ( μ , σ ) (7)
Y/T = 1 ∼ Normal distribution (µ1 , σ1 2 ) (7)
Y/T = 2 ~ Normal distribution ( μ , σ ) (8)
Y/T = 2 ∼ Normal distribution (µ2 , σ2 22) (8)
Y/T = 3 ~Normal distribution ( μ3 , σ3 ) (9)
Y/T = 3 ∼ Normal distribution3CW (µ3 , σ3 2 ) (9)
Y/T = 4 ~ Uniform distribution ( , CW ) (10)
4
3CW
Y/T = 4 ∼ Uniform distribution ( , CW) (10)
where: 4
where:
μ: Mean of the Normal Distribution, initialized with μ =
µ: Mean of the Normal Distribution, initialized with µ0 = CW 8
σ: Variance
Varianceof ofthe
theNormal
NormalDistribution,
Distribution,suchsuchthatthatσ0 σ= = CW . Note that we set a fixed value of σ
σ: 8 . Note that we set a fixed value of σ to
to allow
allow many
many values
values of the
of the back-off
back-off to chosen.
to be be chosen.
AA data
datatype indexTT∈{1,{1,
typeindex 2, 2,
3, 3,
4} 4} is attached
is attached to each
to each packet.
packet. The The back-off
back-off valuesvalues are computed
are computed from
from
the theindex
data data value
index(T),value
and(T), and the distribution
the distribution as shown as shown 6.inAFigure
in Figure truncated6. Anormal
truncated normal
distribution
distribution
is is used
used to ensure to ensureback-off
the positive the positive back-off
can only can only
be positive. Thebe positive.
normal The normal
distribution distribution
allows the node
allows
to pick the node toback-off
a smaller pick a smaller
with high back-off with high
probability. We probability.
fix a smallerWe meanfix atosmaller
draw the mean to draw
small the
back-off
small back-off value. The mean values µ  {µ , µ , µ } are chosen like µ < µ <
value. The mean values µ ∈ {µ1 , µ2 , µ3 } are chosen like µ1 < µ2 < µ3 and there by the back-off values
1 2 3 1 2 µ 3 and there by the
back-off
are BOFF1values
< BOFF2 are
< BOFF1
OFF3<. B
The < BOFF3. The
OFF2uniform uniform on
distribution distribution on[3CW/4,
the interval the interval
CW] [3CW/4,
is usedCW] is used
to draw the
to draw values
back-off the back-off
for thevalues
normal forvehicles,
the normalhencevehicles,
ensuring hence
lower ensuring lower
delays for delays for the
the ambulance thanambulance
the other
than the other
emergency emergency
vehicles vehicles
and normal and normal vehicles.
vehicles.

Figure 6. The
Figure 6. The back-off
back-off distribution
distribution used
used in
in PE-MAC.
PE-MAC.

The mean of the normal distribution is computed using the PE-MAC, Algorithm 1, which is
explained later in this section. The main parameters that we have to talk about are delay threshold
(DThr ), current MAC delay (DK ) and current average MAC delay (DK (Avg)). The delay threshold
(DThr ), which is the maximum allowed delay per node for the transmission of emergency messages.
The delay threshold can be computed by summing the DCF inter frame space time (TDIFS), short
inter frame space time (TSIFS), estimated time to transmit a Request To Send (RTS) frame (TRTS),
estimated time to transmit a Clear To Send (CTS) frame (TCTS), time to transmit a data frame (TData ),
Sensors 2016, 16, 1892 11 of 22

time to transmit an acknowledgement frame (Tack), and error tolerance value for the successful
transmission (et ). The frame structure and the CSMA/CA are depicted in Figure. A node computes
the MAC delay by subtracting the passing time, a packet is passed to the MAC layer from the sent
time, a packet is actually sent onto the link. DK and DK-1 are current MAC delay and stored previous
average MAC delay. “α “ is a constant that determines the effect of the previous average MAC delay
(DK-1 (Avg)) on the current average MAC delay (DK (Avg)). The current average MAC delay can be
written as:
DK (Avg) = (1 − α) × DK + α × DK−1 (Avg) (11)

The PE-MAC algorithm prioritizes the emergency vehicle messages by dividing the contention
window’s interval. The operation of the proposed algorithm is given in the following steps:

• Initialize all the variables.

• Compute the DThr and DK (Avg).
• Compare the DK (Avg) value with the DThr value. If the DK (Avg) ≤ DThr, then the mean value
is assigned with the previous mean value and according to the type of messages, the contention
window interval is divided. This is how a back-off value for the different messages is picked
around the mean value.
• If DK (Avg) > DThr, the mean value is decremented so, that a smaller back-off value will be chosen
in the successive iteration.

The proposed PE-MAC algorithm allows the emergency messages to be intimated to the TMC
within the required and less delay.

Algorithm 1. Priority for Emergency Traffic based MAC (PE-MAC)

1: µ0 ← cw cw
8 , µThr ← 16 , k ← 0 ,
2: Compute DThr and DK (Avg)
3: if (DK (Avg) ≤ DThr ) then
4: µK+1 = µK
5: else
6: µK+1 = µK − 0.5
7: if µK+1 ≤ µThr
8: µK+1 = µ0
9: end if
10: end if
11: CWT1 = [0 , µK+1 + σ ]
12: CWT2 = [µK+1 + σ , µK+1 + 3 σ]
13: CWT3 = [µK+1 + 3 σ , µK+1 + 5 σ]
14: CWT4 = [µK+1 + 5 σ , CW]
This algorithm allows the emergency messages to be sent to the TMC within the required time and
with less delay.

4.2. Simulation Environment

We wanted to examine the possibility of deploying an intelligent and dynamic traffic management
system, which collects the priority information from the vehicles using visual sensing techniques,
delivers the measured information to the TMC with less delay using the PE-MAC protocol and utilizes
this information to effectively manage the traffic signals and thereby the emergency vehicles by
developing a VANET model. A VANET model for the urban traffic management system shown in
Figure 2 is developed and simulated in NS-2. The simulation parameters are given in the Table 3.
Network animator (NAM) of the developed VANET simulation is shown in Figure 7. The black colour
Sensors 2016, 16, 1892 12 of 22

nodes represent the intersection, blue colour nodes represent the traffic signals, cyan colour nodes
represent the normal vehicles, red coloured nodes represent the emergency vehicles, yellow colour
nodes represent the RSUs and mustard colour nodes represent the TMCs (TMCA and TMCB).

Table 3. Simulation parameters.

Parameter Value
Network Area 1500 m × 1500 m
Propagation model Propagation/Two Ray ground
Network interface type Physical/wirelessphy
Sensors 2016, 16, 1892 Interface queue Queue/Droptail/Priqueue 12 of 22
Channel type Channel/Wireless channel
4.2. Simulation Environment Antenna Antenna/OmniAntenna
Visualization tool NAM, Tracing
We wanted to examine Routingthe possibility of deploying an
protocol DSR intelligent and dynamic traffic
management system, whichMCA collects
layerthe priority information from the
IEEE 802.11p vehicles using visual sensing
techniques, delivers theTransmission
measured information
rate to the TMC 9.6 Kbps delay using the PE-MAC
with less
protocol and utilizes this Traffic
information
type to effectively manage the traffic signals and thereby the
CBN
emergency vehicles by developing
Radio delaya VANET model. A VANET 10 m model for the urban traffic
management system shown Link layer type 2 is developed and simulated
in Figure LL in NS-2. The simulation
parameters are given in thePacket
Table 3. size
Network animator (NAM)512 of bytes
the developed VANET simulation
IFQ length
is shown in Figure 7. The black 50
colour nodes represent the intersection, blue colour nodes represent
Initial
the traffic signals, cyan colour energy
nodes represent the normal vehicles,100red
J coloured nodes represent the
emergency vehicles, yellowNo.ofcolour nodes
nodes represent the RSUs and 5 to 100 colour nodes represent the
mustard
Speed 5, 10, 15 and 25 m/s
TMCs (TMCA and TMCB).

Figure 7. NAM of VANET simulation.

Figure 7. NAM of VANET simulation.
Table 3. Simulation parameters.
5. Results and Discussion
Parameter Value
In this section, the performance of the PE-MAC is evaluated
Network Area
and compared with the IEEE802.11p.
1500 m × 1500 m
The three performance metrics: the average
Propagation model
end-to-end delay, throughput
Propagation/Two Ray ground
and residual energy are
evaluated. We have alsoNetwork
compared the average
interface type
end-to-end delay of
Physical/wirelessphy
the proposed PE-MAC with
IEEE 802.11p and IEEE 802.15.4 standards
Interface queue and the existing Enhanced Back-off Selection scheme for
Queue/Droptail/Priqueue
the IEEE 802.15.4 protocol [EBSS] [51].
Channel type We also present the simulation results of the vehicle counting
Channel/Wireless channel
method and distance measurement Antennatechniques. Antenna/OmniAntenna
Visualization tool NAM, Tracing
5.1. Performance of the PE-MAC and IEEE802.11p
Routing protocol DSR
The simulation of the VANET model for the urban traffic
MCA layer management system working with the
IEEE 802.11p
proposed protocol is performed in NS2rate
Transmission with the number of nodes varying from 5 to 100. The simulation
9.6 Kbps
results show the performanceTraffic
of thetype
PE-MAC and IEEE802.11p. CBNThe main parameters to be evaluated
Radio delay 10 m
Link layer type LL
Packet size 512 bytes
IFQ length 50
Initial energy 100 J
 with IEEE 802.11p and IEEE 802.15.4 standards and the existing Enhanced Back-off Selection scheme
for the IEEE 802.15.4 protocol [EBSS] [51]. We also present the simulation results of the vehicle
counting method and distance measurement techniques.

5.1. Performance of the PE-MAC and IEEE802.11p

Sensors 2016, 16,
The1892
simulation of the VANET model for the urban traffic management system working with the 13 of 22
proposed protocol is performed in NS2 with the number of nodes varying from 5 to 100. The
simulation results show the performance of the PE-MAC and IEEE802.11p. The main parameters to
in this simulation
be evaluatedare the simulation
in this average end-to-end delay,
are the average throughput
end-to-end delay, and residual
throughput andenergy.
residualThe histogram
energy.
plotted The
in Figure
histogram8 plotted
compares the 8average
in Figure comparesend-to-end delay ofdelay
the average end-to-end the of
proposed PE-MAC
the proposed PE-MACand the
and the protocol
IEEE 802.11p, IEEE 802.11p,
underprotocol under
varying varyingof
number number
node of node conditions.
conditions.

120 PE-MAC IEEE 802.11p

Average end-to-end delay (Sec) 100

80

60

40

20

0
10 20 30 40 50 60 70 80 90 100
Number of nodes

Figure 8. Impact
Figure of number
8. Impact of nodes
of number on average
of nodes end-to-end
on average delay:
end-to-end proposed
delay: PE-MAC
proposed PE-MACvsvsIEEE
IEEE802.11p.
802.11p.

The sizeThe of size

the network or theornumber
of the network of nodes
the number of nodesshows
showsitsits
impact
impacton onthe
theaverage end-to-end delay.
average end-to-end
The end-to-end
delay. Thedelay increases
end-to-end delaywith the node
increases number.
with the DespiteDespite
node number. this increase, it can
this increase, be clearly
it can be clearlyobserved
from thisobserved
figure from this proposed
that the figure that the proposed
PE-MAC PE-MACa achieves
achieves a lower end-to-end
lower end-to-end delay compared
delay compared to the IEEE
802.11p.toWhen
the IEEE 802.11p.
there When
are 10, 50 there
and are
10010, 50 and
nodes in100
thenodes in the the
network, network,
PE-MACthe PE-MAC
confirmsconfirms the
the end-to-end
end-to-end delay of 15 s, 18 s and 40 s respectively, and the IEEE 802.11p achieves end-to-end delay
delay of 15 s, 18 s and 40 s respectively, and the IEEE 802.11p achieves end-to-end delay of 17, 28 and
of 17, 28 and 98 s respectively. The achieved improvement here is almost 11% at the number of nodes
98 s respectively.
equal to 10The achievedtoimprovement
and increases about 60% when here
the is almost
number of11%
nodes atequal
the number of nodes
to 100. Thereby, weequal
confirm to 10 and
increases to the
that about 60% when
proposed PE-MACthe number
delivers of
thenodes equal faster
information to 100.than
Thereby, we confirm
the standard that the proposed
IEEE 802.11p.The
PE-MAC delivers
histogram the information
plotted faster than
in Figure 9 compares the standard
the average end-to-end IEEE
delay802.11p.The
of the proposed histogram
PE-MAC and plotted in
Figure 9the IEEE 802.11p,
compares under various
the average periodicdelay
end-to-end packetoftransmission
the proposed intervals.
PE-MACThis figure clearly
and the IEEEshows that under
802.11p,
various the proposed
periodic PE-MAC
packet achieves lower
transmission end-to-end
intervals. Thisdelay compared
figure clearlytoshows
the standard IEEEproposed
that the 802.11p. The PE-MAC
achieved improvement here is almost 28% in the case of the transmission interval of 10 s, and
achieves lower end-to-end delay compared to the standard IEEE 802.11p. The achieved improvement
increases to about 42% when the transmission interval is 60 s.
here is almost 28% in the case of the transmission interval of 10 s, and increases to about 42% when the
transmission interval
Sensors 2016, 16, 1892is 60 s. 14 of 22

20
IEEE 802.11p PE-MAC
Avg end-end delay (Sec)

15

10

0
10 20 30 60
Inter Packet transmission interval (Sec)

Figure 9. Impact of inter packet transmission interval on average end-to-end delay: proposed PE-
Figure 9. Impact of inter packet transmission interval on average end-to-end delay: proposed PE-MAC
MAC vs IEEE 802.11p.
vs IEEE 802.11p.
Figure 10 shows the impact of the number of nodes on residual energy for both the proposed
PE-MAC and IEEE 802.11p. The PE-MAC outperforms the IEEE 802.11p in terms of energy
consumption. The residual energy of the PE-MAC is more compared to the IEEE 802.11p. As the node
number increases, the intermediate nodes share the work and less energy is consumed for the
transmission. These results have shown that the proposed scheme consumes less energy compared
to the standard IEEE 802.11p. The initial energy we have taken is 100 J. The residual energy of the PE-
MAC at the number of nodes equal to 10 is 98 J that means the energy consumed by PE-MAC is 2 J.
The energy consumed by PE-MAC at 10, 50 and 100 number of nodes is 2, 5 and 8 J, respectively, and
 Avg e
5

0
10 20 30 60
Sensors 2016, 16, 1892 Inter Packet transmission interval (Sec) 14 of 22

Figure 9. Impact of inter packet transmission interval on average end-to-end delay: proposed PE-
Figure MAC
10 shows
vs IEEEthe impact of the number of nodes on residual energy for both the proposed
802.11p.
PE-MAC and IEEE 802.11p. The PE-MAC outperforms the IEEE 802.11p in terms of energy
consumption. Figure
The10residual
shows theenergy
impact of
of the
thenumber
PE-MAC of nodes on residual
is more compared energy
tofor
theboth the 802.11p.
IEEE proposed As the
PE-MAC and IEEE 802.11p. The PE-MAC outperforms the IEEE 802.11p in terms of energy
node number increases, the intermediate nodes share the work and less energy is consumed for the
consumption. The residual energy of the PE-MAC is more compared to the IEEE 802.11p. As the node
transmission.
numberThese results
increases, have shownnodes
the intermediate that theshareproposed
the work scheme
and less consumes less energy
energy is consumed for compared
the
to the standard IEEE
transmission. 802.11p.
These results The
haveinitial
shown energy we havescheme
that the proposed taken is 100 J. The
consumes residual
less energy energy of the
compared
PE-MAC to at
thethe number
standard of802.11p.
IEEE nodes equal to 10
The initial is 98we
energy J that
havemeans
taken isthe
100energy consumed
J. The residual byofPE-MAC
energy the PE- is 2 J.
MAC at the number of nodes equal to 10 is 98 J that means the energy consumed
The energy consumed by PE-MAC at 10, 50 and 100 number of nodes is 2, 5 and 8 J, respectively, by PE-MAC is 2 J. and
The energy consumed by PE-MAC at 10, 50 and 100 number of nodes is 2, 5 and 8 J, respectively, and
the energy consumed by IEEE 802.11p at 10, 50 and 100 number of nodes is 5, 8 and 16 J, respectively.
the energy consumed by IEEE 802.11p at 10, 50 and 100 number of nodes is 5, 8 and 16 J, respectively.
The energy consumed by the proposed PE-MAC is almost 50% less than that of the energy consumed
The energy consumed by the proposed PE-MAC is almost 50% less than that of the energy consumed
by the IEEE
by the802.11p protocol.
IEEE 802.11p protocol.

100 PE-MAC IEEE 802.11p

95
Residual Energy (J)

90

85

80

75
10 20 30 40 50 60 70 80 90 100
Number of Nodes

Figure 10. Impact of number of nodes on residual energy: proposed PE-MAC vs IEEE 802.11p.
Figure 10. Impact of number of nodes on residual energy: proposed PE-MAC vs IEEE 802.11p.
Figure 11 shows the impact of the number of nodes on the throughput for both the IEEE 802.11p
standard
Figure and the
11 shows proposed
the impact PE-MAC. In comparison
of the number of nodeswith
on IEEE 802.11p, PE-MAC
the throughput keeps
for both thea IEEE
better802.11p
throughput for both the lesser and higher node number. When there are
standard and the proposed PE-MAC. In comparison with IEEE 802.11p, PE-MAC keeps 20, 50 and 100 nodes in thea better
network, the PE-MAC confirms the positive data delivery of about 7, 9.5 and 15.8 Kbps, respectively,
throughput for both the lesser and higher node number. When there are 20, 50 and 100 nodes in the
network, the 2016,
Sensors PE-MAC
16, 1892 confirms the positive data delivery of about 7, 9.5 and 15.8 Kbps,15 respectively,
of 22
and the positive data delivery of the IEEE 802.11p is 3.3, 4.2 and 5 Kbps, respectively. The achieved
and the positive data delivery of the IEEE 802.11p is 3.3, 4.2 and 5 Kbps, respectively. The achieved
improvement here is almost 52% in the case the number of nodes equals 20, and increases to about
improvement here is almost 52% in the case the number of nodes equals 20, and increases to about
70% when the number of nodes equals 100.
70% when the number of nodes equals 100.

20
PE-MAC IEEE 802.11p

16
Throughput (Kbps)

12

0
10 20 30 40 50 60 70 80 90 100
Number of Nodes

Figure 11. Impact of the number of nodes on throughput: proposed PE-MAC vs IEEE 802.11p.
Figure 11. Impact of the number of nodes on throughput: proposed PE-MAC vs IEEE 802.11p.
Figure 12 shows the impact of the network size on the average end-to-end delay of all the data
type messages. We run our simulation with the PE-MAC and the results depicted in this figure clearly
show that the end-to-end delay of the ambulance data is very small compared to the other emergency
vehicles and normal vehicles. From these results, we can also observe that the average end-to-end
delay increases with the network size. The results have proven that the proposed PE-MAC delivers
the emergency messages to the TMC with less delay.
 Thr
4

0
10 20 30 40 50 60 70 80 90 100
Sensors 2016, 16, 1892 Number of Nodes 15 of 22

Figure 11. Impact of the number of nodes on throughput: proposed PE-MAC vs IEEE 802.11p.
Figure 12 shows the impact of the network size on the average end-to-end delay of all the data
Figure
type messages. We12run
shows
ourthe impact of the
simulation withnetwork size on the
the PE-MAC andaverage end-to-end
the results delay in
depicted of all
thisthefigure
data clearly
type messages. We run our simulation with the PE-MAC and the results depicted in this figure clearly
show that the end-to-end delay of the ambulance data is very small compared to the other emergency
show that the end-to-end delay of the ambulance data is very small compared to the other emergency
vehiclesvehicles
and normal vehicles.
and normal From
vehicles. Fromthese
theseresults, wecan
results, we canalso
also observe
observe thatthat the average
the average end-to-end
end-to-end
delay increases
delay increases with the network size. The results have proven that the proposed PE-MAC deliversdelivers
with the network size. The results have proven that the proposed PE-MAC
the emergency messages
the emergency to the
messages TMC
to the TMCwithwithless
lessdelay.
delay.

Ambulances FireFighters Police Vehicles Normal Vehicles

100
Avg end-to-end delay (Sec)

80

60

40

20

0
10 20 30 40 50 60 70 80 90 100
Nunber of Nodes

Figure Figure 12. Impact

12. Impact of number
of number of of nodeson
nodes onaverage
average end-to-end
end-to-enddelay: all data
delay: type messages.
all data type messages.

The histogram plotted in Figure 13 compares the average end-to-end delay of the proposed PE-
TheMAChistogram plotted
with standard IEEEin Figure
802.11p, 13802.15.4
IEEE compares theexisting
and the average end-to-end
Enhanced Back-offdelay of scheme
Selection the proposed
PE-MAC forwith standard
the IEEE 802.15.4IEEE 802.11p,
protocol IEEE
[EBSS, an 802.15.4
existing and
protocol to the existing
ensure Enhanced
fast transmission of Back-off Selection
the detected
schemeevents
for the on IEEE
the road towardsprotocol
802.15.4 the TMC][EBSS,
[52], foran
various periodic
existing packet to
protocol transmission
ensure fastintervals. The
transmission of
achieved
the detected improvement
events here istowards
on the road almost 31%theover IEEE[52],
TMC] 802.15.4, 11% over periodic
for various EBSS and packet
28% overtransmission
IEEE
802.11p in the case of the transmission interval of 5 s, and increases to about 45% over IEEE 802.15.4,
intervals. The achieved improvement here is almost 31% over IEEE 802.15.4, 11% over EBSS and 28%
over IEEE 802.11p
Sensors in the case of the transmission interval of 5 s, and increases to about
2016, 16, 1892 16 of45%
22 over
IEEE 802.15.4, 16.6% over EBSS and 42% over IEEE 802.11p when the transmission interval is 60 s.
16.6% over EBSS and 42% over IEEE 802.11p when the transmission interval is 60 s. The results have
The results have proven that the proposed PE-MAC achieves lower end-to-end delay compared to the
proven that the proposed PE-MAC achieves lower end-to-end delay compared to the other schemes
other schemes
considered,considered,
existing andexisting and standard.
standard.

PE-MAC IEEE 802.11p EBSS IEEE 802.15.4

Inter-packet transmission interval (Sec)

60

30

10

0 5 10 15 20
Avg end-to-end delay (Sec)

Figure 13. Impact of inter packet transmission interval on average end-to-end delay: for proposed PE-
Figure 13. Impact of inter packet transmission interval on average end-to-end delay: for proposed
MAC, EBSS, standard IEEE 802.15.4 and IEEE 802.11p.
PE-MAC, EBSS, standard IEEE 802.15.4 and IEEE 802.11p.
5.2. Simulation Results of Vehicle Counting and Distance Measurement Techniques
In this subsection, we want to give the complete description of using visual sensing techniques
to vehicle detection and counting. We collected the videos of road traffic with emergency vehicles
using a digital camera setup. Five input videos of road traffic, around four to eight minutes were
recorded. The spatial resolution of the recorded videos was 1920 × 1080 pixels, with temporal
Sensors 2016, 16, 1892 16 of 22

5.2. Simulation Results of Vehicle Counting and Distance Measurement Techniques

In this subsection, we want to give the complete description of using visual sensing techniques to
vehicle detection and counting. We collected the videos of road traffic with emergency vehicles using a
digital camera setup. Five input videos of road traffic, around four to eight minutes were recorded.
The spatial resolution of the recorded videos was 1920 × 1080 pixels, with temporal resolution of
30 frames per second. The vehicle counting method was implemented in MATLAB to count the
vehicles moving along with the emergency vehicle towards next intersection.
To achieve better results, the vehicle detection process should be done in the grayscale domain.
Firstly a grayscale conversion is performed on each RGB video frame. The RGB frame and grayscale
frame are shown in Figure 14a,b, respectively. Subsequently, we do foreground extraction by using the
frame differencing technique. The difference image is shown in Figure 14c. The following step is to
define a detection region. The subsequent step is to do thresholding, a method of image segmentation
which separates
Sensors the foreground pixels from the background pixels.
2016, 16, 1892 17 of 22

Figure
Figure 14. 14. Emergency
Emergency vehicledetection
vehicle detectionprocess:
process: (a)
(a) RGB
RGB frame;
frame;(b)(b)Grayscale
Grayscaleframe; (c) (c)
frame; Difference
Difference
image; (d) Binary image; (e) Dilated image; (f) Hole filling image; (g) Eroded image; (h) Tagged Vehicle.
image; (d) Binary image; (e) Dilated image; (f) Hole filling image; (g) Eroded image; (h) Tagged Vehicle.

The thresholding operation results in binary image as shown in Figure 14d, which contains
both false and missing foreground pixels. Hence, some more morphological techniques are used to
eradicate noise and improve foreground objects. The morphological dilation is performed to recover
some missing foreground pixels. The dilated image is shown in Figure 14e. The hole filling operation
Sensors 2016, 16, 1892 17 of 22

is performed to suppress black pixels surrounded by white regions. The resulting image after the hole
filling operation is shown in Figure 14f. The morphological erosion operation will further convert the
foreground items into the background. The result after a morphological erosion operation is shown
in Figure 14g. After the morphological operations, the foreground items are found to be smoother
and clearer. The detection zone defined previously is used to track and count the vehicles. When a
vehicle enters the detection region, the bounding box appears around the moving vehicle as shown in
Figure 14h.
Distance measurement plays an important role in determining the green light sequence. Distance
measurement techniques help us find the nearest emergency vehicle to the intersection. We have
measured the distance between the emergency vehicle and intersection using the Euclidean distance,
Manhattan distance and Canberra distance techniques in MATLAB.
Results of distance measurement using the Euclidean distance, Manhattan distance and Canberra
distance techniques are shown in Figure 15a–c, respectively. We have done distance measurements
at several
Figurepoints. The simulation
14. Emergency values are
vehicle detection compared
process: (a) RGBwith the(b)
frame; true distance
Grayscale measured
frame; practically.
(c) Difference
The accuracy of each measurement technique is given in Table 4.
image; (d) Binary image; (e) Dilated image; (f) Hole filling image; (g) Eroded image; (h) Tagged Vehicle.

Figure 15. Distance

Figure 15. Distance measurement
measurement techniques:
techniques: (a)
(a) Euclidean
Euclidean distance;
distance; (b)
(b) Manhattan
Manhattan distance;
distance;
(c) Canberra distance.
(c) Canberra distance.

Table 4. Experiment Results.

Distance Measurement at Discrete Points

Distance (All Distances Are in Meters)
Measurement Accuracy Outcome
P1 P2 P3
Techniques
True Value: 142 True Value :121 True Value: 62
The simulation values
Euclidean
140.03 120.25 60.66 98.60% are always very nearer to
Distance
true values.
Only at some points, the
Manhattan
138 54.03 56.45 77.61% simulation values are
Distance
nearer to true values.
The simulation values
Canberra
45.66 28 19.25 28.78% are always distant from
Distance
the true values.

The simulation results show that the Euclidean distance outperforms other distance measurement
techniques. Figure 16 gives the comparison between the distance measurement techniques.
 45.66 28 19.25 28.78% are always distant
Distance
from the true values.

The simulation results show that the Euclidean distance outperforms other distance
measurement techniques. Figure 16 gives the comparison between the distance measurement
Sensors 2016, 16, 1892 18 of 22
techniques.

Figure 16. Comparison between the distance measurement techniques.

After obtaining
After obtaining the
the Euclidean
Euclidean distance,
distance, we
we measured
measured the
the speed
speed ofof the
the emergency
emergency vehicle
vehicle (using
(using
∆d/∆t) and counted the vehicles moving along with the emergency vehicle towards
∆d/∆t) and counted the vehicles moving along with the emergency vehicle towards next intersection. next intersection.
The measured
The measuredvaluesvaluesofofvehicle
vehicle count,
count, Euclidean
Euclidean distance
distance andand speed
speed are shown
are shown in Figure
in Figure 17. The17. The
traffic
traffic management
management centercenter can utilize
can utilize this this information
information in aintraffic
a traffic signalcontrol
signal controlmodule,
module,resulting
resulting in in
an efficient emergency traffic management process. All the existing works depend
an efficient emergency traffic management process. All the existing works depend on some kind on some kind of
infrastructure
of infrastructure andand
require extra
require costcost
extra equipment. OurOur
equipment. scheme
schemeutilizes ultrasonic
utilizes sensors,
ultrasonic RSUsRSUs
sensors, and
existing surveillance cameras. The image processing-based approach cuts the
and existing surveillance cameras. The image processing-based approach cuts the installation and installation and
maintenance
Sensors costs compared
2016, 16, costs
maintenance 1892 compared to to existing
existing emergency
emergency vehicle
vehicle pre-emption
pre-emption (EVP)
(EVP) systems
systems[52,53].
[52,53].19 of 22

Figure 17. Measured data.

6. Conclusions
6. Conclusions
This paper
This paper has
has presented
presented anan approach
approach toto schedule
schedule emergency
emergency vehicles
vehicles in
in traffic.
traffic. The
The approach
approach
combines the measurement of distance between the emergency vehicle and intersections visual
combines the measurement of distance between the emergency vehicle and intersections using using
sensingsensing
visual methods, vehicle vehicle
methods, counting and timeand
counting sensitive alert transmission
time sensitive within the
alert transmission sensorthe
within network.
sensor
The distance
network. Thebetween
distancethebetween
emergency thevehicle and the
emergency intersection
vehicle and theis calculated from
intersection is visual data using
calculated from
Euclidean
visual datadistance, Manhattan
using Euclidean distance
distance, and Canberra
Manhattan distance
distance techniques
and Canberra for comparison.
distance techniquesThefor
experimental The
comparison. results have shown
experimental resultsthat
havethe
shownEuclidean
that thedistance
Euclideanoutperforms other distance
distance outperforms other
measurement
distance techniques
measurement and is suitable
techniques for real-time
and is suitable applications.
for real-time A complete
applications. description
A complete of the use
description of
of visual sensing techniques in vehicle detection and counting is also presented. The measured
information like vehicle count, distance and speed are very useful for a traffic management center to
manage emergency traffic efficiently. After obtaining the measured information, how fast it is
delivered to the TMC is very important. For that, we have proposed a PE-MAC protocol, which is a
new back off selection and contention window adjustment scheme to achieve low broadcast delay
Sensors 2016, 16, 1892 19 of 22

the use of visual sensing techniques in vehicle detection and counting is also presented. The measured
information like vehicle count, distance and speed are very useful for a traffic management center
to manage emergency traffic efficiently. After obtaining the measured information, how fast it is
delivered to the TMC is very important. For that, we have proposed a PE-MAC protocol, which
is a new back off selection and contention window adjustment scheme to achieve low broadcast
delay for emergency messages. A VANET model for the urban traffic management system used in
this work is developed and simulated in NS-2. The NS-2 simulation results have shown that the
PE-MAC outperforms the IEEE 802.11p in terms of average end-to-end delay, throughput and energy
consumption. The performance evaluation results have proven that the proposed PE-MAC prioritizes
the emergency vehicle data and delivers the emergency messages to the TMC with less delay compared
to the IEEE 802.11p. The transmission delay of the proposed PE-MAC is also compared with the
standard IEEE 802.15.4, and EBSS and the comparative results have proven the effectiveness of the
PE-MAC over them. The improvement achieved by PE-MAC is almost 31% over IEEE 802.15.4, 11%
over EBSS and 28% over IEEE 802.11p in the case of a transmission interval of 5 s, and increases to
about 45% over IEEE 802.15.4, 16.6% over EBSS and 42% over IEEE 802.11p when the transmission
interval is 60 s. The results have proven that the proposed PE-MAC achieves lower end-to-end delay
compared to the considered schemes. From our work, we can confirm that the emergency vehicle
information is measured accurately by using visual sensing methods and the measured information is
delivered to the TMC in less time by adopting the proposed PE-MAC protocol. The potential impact
of the proposed work is extensive as, being an inexpensive, effective, and accurate, it can be effectively
applied in practice. Further research should be done on distance measurement in bad weather and
high traffic conditions.

Author Contributions: All the authors have equally contributed to the development and writing of this article.
Conflicts of Interest: The authors declare no conflict of interest.

References
1. Djahel, S.; Smith, N.; Wang, S.; Murphy, J. Reducing emergency services response time in smart cities:
An advanced adaptive and fuzzy approach. In Proceedings of the IEEE First International Smart Cities
Conference (ISC2), Guadalajara, Mexico, 25–28 October 2015; pp. 1–8.
2. National Highway Traffic Safety Administration (NHTSA) and the Economic and Social Impact of Motor
Vehicle Crashes. Available online: https://crashstats.nhtsa.dot.gov/Api/Public/ViewPublication/812013
(accessed on 14 December 2015).
3. Statistics on Emergency Vehicle Accidents in the U.S. Available online: http://www.arnolditkin.
com/Personal-Injury-Blog/2015/November/Statistics-on-Emergency-Vehicle-Accidents-in-the.aspx
(accessed on 5 December 2015).
4. Nellore, K.; Hancke, G.P. A survey on urban traffic management system using wireless sensor networks.
Sensors 2016, 16, 157. [CrossRef] [PubMed]
5. Rajeshwari, S.; Santhoshs, H.; Varaprasad, G. Implementing intelligent traffic control system for congestion
control, ambulance clearance and stolen vehicle detection. IEEE Sens. J. 2015, 15, 1109–1113.
6. Sireesha, E.; Rakesh, D. Intelligent traffic light system to prioritized emergency purpose vehicles based on
wireless sensor network. Int. J. Res. Stud. Sci. Eng. Technol. 2014, 1, 24–27.
7. Shruthi, K.R.; Vinodha, K. Priority based traffic lights controller using wireless sensor networks. Int. J.
Electron. Signal Syst. 2012, 1, 58–61.
8. Hussian, R.; Sandhy, S.; Vinita, S.; Sandhya, S. WSN applications: Automated intelligent traffic control
system using sensors. Int. J. Soft Comput. Eng. 2013, 3, 77–81.
9. Nabeel, M.M.; EI-Dien, M.F.; EI-Kader, S.A. Intelligent vehicle recognition based on wireless sensor network.
Int. J. Comput. Sci. Issues 2013, 10, 164–174.
10. Nellore, K.; Melingi, S.B. Automatic Traffic Monitoring System Using Lane Centre Edges. IOSR J. Eng. 2012,
2, 1–8.
Sensors 2016, 16, 1892 20 of 22

11. Uddin, S.M.; Das, K.A.; Taleb, A.M. Real-time area based traffic density estimation by image processing
for traffic signal control: Bangladesh perspective. In Proceedings of the IEEE International Conference
on Electrical Engineering and Information Communication Technology (ICEEICT), Dhaka, Bangladesh,
21–23 May 2015; pp. 1–5.
12. Traffic Signal Preemption for Emergency Vehicles. Available online: http://ntl.bts.gov/lib/jpodocs/repts_
te/14097_files/14097.pdf (accessed on 5 December 2015).
13. Farheena, S.; Chandak, B.M. An approach towards traffic management system using density calculation and
emergency vehicle alert. IOSR J. Comput. Sci. 2014, 4, 24–27.
14. Bharadwaj, R.; Deepak, J.; Baranitharam, M.; Vaidehi, V.V. Efficient dynamic traffic control system using
wireless sensor networks. In Proceedings of the IEEE International Conference on Recent Trends in
Information Technology (ICRTIT), Chennai, India, 25–27 July 2013; pp. 668–673.
15. Mithun, C.N.; Rashid, U.N.; Rahman, M.M.S. Detection and classification of vehicles from video using
multiple Time-Spatial images. IEEE Trans. Intell. Transp. Syst. 2012, 13, 1215–1225. [CrossRef]
16. Kanungo, T.; Mount, M.D.; Netanyahu, S.N.; Piatko, D.C.; Silverman, R.; Wu, Y.A. An efficient K-means
clustering algorithm: analysis and implementation. IEEE Trans. Pattern Anal. Mach. Intell. 2002, 24, 881–892.
[CrossRef]
17. Rahaman, O.Md.; Hong, S.C.; Lee, S.; Bang, C.Y. ATLAS: A traffic load aware sensor MCA design for
Collaborative body area sensor networks. Sensors 2011, 11, 11560–11580. [CrossRef] [PubMed]
18. Abubakr, S.E.; Halla, O.A.; Tahani, A.A. A GPS based traffic light pre-emption control system for emergency
vehicles. In Proceedings of the IEEE International conference on Computing, Electrical and Electronics
Engineering, Khartoum, Sudan, 26–28 August 2013; pp. 724–729.
19. Nellore, K.; Palepu, V.S.; Palepu, M.R.D.; Chenchela, V.K. Improving the lifespan of wireless sensor network
via efficient carrier sensing scheme-CSMA/SDF. Int. J. Eng. Sci. Res. Technol. 2016, 5, 723–732.
20. Abu-Mahfouz, A.M.; Hancke, G.P. ns-2 Extension to Simulate Localization Systems in Wireless Sensor
Networks. In Proceedings of IEEE Africon 2011, Livingstone, Zambia, 13–15 September 2011; pp. 1–7.
21. Silva, B.; Hancke, G.P. IR-UWB based Non-line-of-sight Identification in Harsh Environments: Principles
and Challenges. IEEE Trans. Ind. Inform. 2016, 12, 1188–1195. [CrossRef]
22. Adnan, M.A.; Hancke, G.P. An Efficient Distributed Localization Algorithm for Wireless Sensor Networks:
Based on Smart References Selection Criteria. Int. J. Sens. Netw. 2013, 13, 94–111.
23. Fan, K.; Chen, J.; Cao, Q. Design and research on traffic of wireless sensor network based on LabVIEW.
In Proceedings of the 2nd International Symposium on Computer, Communication, Control and Automation
(3CA 2013), Singapore, 1–2 December 2013; pp. 6–9.
24. AI-Ostath, N.; Selityn, F.; AI-Roudhan, Z.; EI-Abd, M. Implementation of an emergency vehicle to traffic
lights communication system. In Proceedings of the 7th International Conference on New Technologies,
Mobility and Security (NTMS), Paris, France, 27–29 July 2015; pp. 1–5.
25. Koonce, P.; Lee, R.; Kevin, L.; Shaun, Q.; Scott, B.; Cade, B.; Jim, B.; Phil, T.; Tom, U. Signal Timing
Manual. Available online: http://nacto.org/docs/usdg/signal_timing_manual_fhwa.pdf (accessed on
26 March 2016).
26. Chinnappen-Rimer, S.; Hancke, G.P. Actor coordination using info-gap decision theory in wireless sensor
and actor networks. Int. J. Sens. Netw. 2011, 10, 177–191. [CrossRef]
27. Mittal, K.A.; Bhandari, D. A novel approach to implement green wave system and detection of stolen Vehicles.
In Proceedings of the 2013 IEEE 3rd International Advance Computing Conference (IACC), Ghaziabad,
India, 22–23 February 2013; pp. 1055–1059.
28. Varaprasad, G. High stable power aware multicast algorithm for mobile ad hoc networks. IEEE Sens. J. 2013,
1, 1442–1446. [CrossRef]
29. Boris, S.K. The physics of green-wave breakdown in a city. Europhys. Lett. 2013, 102, 28010.
30. Sundar, R.; Hebbar, S.; Golla, V. Implementing intelligent traffic control system for congestion control,
Ambulance clearance, and Stolen vehicle detection. IEEE Sens. J. 2015, 15, 1109–1113.
31. Hegde, R.; Sail, R.R.; Indira, S.M. RFID and GPS based automatic lane clearance system for ambulance. Int. J.
Adv. Elect. Electron. Eng. 2013, 2, 102–107.
Sensors 2016, 16, 1892 21 of 22

32. Jayaraj, V.; Hemanth, C. Emergency vehicle signaling using VANETS. In Proceedings of the IEEE 17th
International Conference on High Performance Computing and Communications (HPCC), 2015 IEEE 7th
International Symposium on Cyberspace safety and Security (CSS), and 2015 IEEE 12th International
Conference on Embedded Software and Systems (ICESS), New York, NY, USA, 24–26 August 2015;
pp. 734–739.
33. Chenchela, V.K.; Palepu, V.S.; Nellore, K.; Palepu, M.R.D. Improving the quality of Service (QOS) of
Connection Admission control Mechanism(CAC) using two dimensional queuing model. IOSR J. Electr.
Electron. Eng. 2016, 11, 17–27.
34. Collotta, M.; Bello, L.L.; Pau, G. A novel approach for dynamic traffic lights management based on wireless
sensor networks and multiple fuzzy logic controllers. Expert Syst. Appl. 2015, 42, 5403–5415. [CrossRef]
35. Hancke, P.G.; De Silva, B.; Hancke, G.P. The Role of Advanced Sensing in Smart Cities. Sensors 2013, 13,
393–425. [CrossRef] [PubMed]
36. Potter, C.; Hancke, G.P.; de Silva, B. Machine-to-Machine: Possible Applications in Industrial Networks.
In Proceedings of the IEEE International Conference on Industrial Technology (ICIT), Cape Town,
South Africa, 25–28 February 2013; pp. 1321–1326.
37. Kruger, P.C.; Hancke, G.P. Benchmarking Internet of things devices. In Proceedings of the IEEE International
Conference on Industrial Informatics, Porto Alegre, Brazil, 27–30 July 2014; pp. 611–616.
38. Kruger, P.C.; Hancke, G.P. Implementing the Internet of Things vision in industrial wireless sensor networks.
In Proceedings of the IEEE International Conference on Industrial Informatics, Porto Alegre, Brazil,
27–30 July 2014; pp. 627–632.
39. Chiwewe, T.; Mbuya, C.; Hancke, G.P. Using Cognitive Radio for Interference-Resistant Industrial Wireless
Networks: An Overview. IEEE Trans. Ind. Inf. 2015, 11, 1466–1481.
40. Kumar, A.; Hancke, G.P. An Energy-Efficient Smart Comfort Sensing System Based on the IEEE 1451 Standard
for Green Buildings. IEEE Sens. J. 2014, 14, 4245–4252. [CrossRef]
41. Cheng, B.; Cui, L.; Jia, W.; Zhao, W.; Hancke, G.P. Multiple Regions of Interest Coverage in Camera Sensor
Networks for Tele-Intensive Care Units. IEEE Trans. Ind. Inf. 2016. [CrossRef]
42. Rohini Priya, P.; Anju Joy, J.; Sumathy, G. Traffic light pre-emption control system for emergency vehicles.
SSRG Int. J. Electron. Commun. Eng. 2015, 2, 21–25.
43. Dobre, A.R.; Nita, A.V.; Ciobanu, A.; Negrescu, C.; Stanomir, D. Low computational methods for siren
detection. In Proceedings of the IEEE 21st International Symposium for Design and Technology in Electronic
packaging (SIITME), Brasov, Romania, 22–25 October 2015; pp. 291–295.
44. Miyazaki, T.; Kitazono, Y.; Shimakawa, M. Ambulance siren detection using FFT and dsPIC. In Proceedings
of the First IEEE/IIAE International Conference on Intelligent System and Image processing, Kitakyushu,
Japan, 26–27 September 2013; pp. 266–269.
45. Meucci, F.; Pierucci, L.; del Re, E.; Lastrucci, L.; Desii, P. Areal-time siren detection to improve safety of
guide in traffic environment. In Proceedings of the IEEE 16th International Conference on European Signal
Processing, Lausanne, Switzerland, 25–29 August 2008; pp. 1–5.
46. Eisenson, H.; Ross, J.M. Device and System to Alert Vehicles and Pedestrians of Approaching Emergency
Vehicles and Emergency Situations, United Sates Patent Application Publications. Available online:
https://docs.google.com/viewer?url=patentimages.storage.googleapis.com/pdfs/US20130049985.pdf
(accessed on 20 February 2015).
47. Fazenda, B.; Atmoko, H.; Gu, F.; Guan, L.; Ball, A. Acoustic based safety emergency vehicle detection for
intelligent transport systems. In Proceedings of the IEEE International Conference ICROS-SICE, Fukuoka,
Japan, 18–21 August 2009; pp. 4250–4255.
48. David, A.; Stefan, L.; Marc, F.; Dieter, K. Emergency Vehicle Detection with Digital Image Sensor. Available
online: https://docs.google.com/viewer?url=patentimages.storage.googleapis.com/pdfs/US20150161458.
pdf (accessed on 20 February 2015).
49. Malkauthekar, D.M. Analysis of Euclidean distance and Manhattan distance measure in Face recognition.
In Proceedings of the IET 3rd International Conference on Computational Intelligence and Information
Technology (CIIT), Herts, UK, 18–19 October 2013; pp. 503–507.
50. Srimani, K.P.; Mahesh, S.; Bhyratae, A.S. Improvement of Traditional K-means algorithm through the
regulation of distance metric parameters. In Proceedings of the IEEE 7th International Conference on
Intelligent System and Control (ISCO), Coimbatore, India, 4–5 January 2013; pp. 393–398.
Sensors 2016, 16, 1892 22 of 22

51. Horiya, I.B.; Soufiene, D.; John, M. Improving emergency messages transmission delay in road monitoring
based WSNs. In Proceedings of the IEEE Joint IFIP Wireless and Mobile Networking Conference (WMNC),
Dubai, United Arab Emirates, 23–25 April 2013; pp. 1–8.
52. Wantanee, V.; Ozan, K.T. Priority Management of Emergency Vehicles at Intersections Using Self Organized
Traffic Control. In Proceedings of the IEEE Conference on Vehicular Technology (VTC Fall), Quebec, QC,
Canada, 3–6 September 2012; pp. 1–4.
53. Chen, C.Y.; Chen, P.Y.; Chen, W.T. A Novel Emergency Vehicle Dispatching System. In Proceedings of the
IEEE 77th Conference on Vehicular Technology, Dresden, Germany, 2–5 June 2013; pp. 1–5.

© 2016 by the authors; licensee MDPI, Basel, Switzerland. This article is an open access
article distributed under the terms and conditions of the Creative Commons Attribution
(CC-BY) license (http://creativecommons.org/licenses/by/4.0/).